Predictive analysis allows next-generation cyber defense that is more proactive than current approaches based solely on intrusion detection. In this talk, we will discuss various approaches to predicting and projecting cyber attacks. Graph-based models are dominating the field since the foundation of this research area. Attack graphs were used to traverse through the attacker’s actions and project the continuation of an ongoing attack. Later, attack graphs were combined with Bayesian networks and Markov models to reflect the probabilistic nature of predictions and overcome uncertainties in observation of attack steps. However, there are still open issues, such as how to create such models and evaluate the predictions. The talk will shed light on using graphs in this research area and summarize resolved and open issues.